faraz7530/physical-ai-textbook-api

Add textbook info to prompts, prioritize Groq for chat

Add LLM provider logging and fix fallback logic

Switch to Groq for chat, keep Gemini for embeddings, fix score threshold to 0.3

Add debug logging for Qdrant score filtering

Trigger rebuild - fix Qdrant query_points and score threshold

Fix: lower retriever score threshold to 0.3 to match Gemini embedding scores

Lower score threshold to 0.3 and add debug logging

Fix: use query_points with manual score filtering for AsyncQdrantClient

Fix: use correct search() method for AsyncQdrantClient with score_threshold

Fix Qdrant search: use search() instead of query_points() for proper score filtering

Use gemini-2.0-flash - correct model name for OpenAI endpoint

Switch to gemini-1.5-flash for better rate limits

Fix Qdrant search threshold from 0.7 to 0.5 for better retrieval

fix: lower RAG score threshold from 0.7 to 0.5 for better recall, add LiDAR synonyms

Update backend with latest changes

Fix README.md for HF Space config

Fix expires_at datetime conversion

Fix datetime to string conversion for Better Auth response

Add Better Auth compatible endpoints for signup and signin

Fix vector dimension for Gemini embeddings, add translation service

Fix vector dimension for Gemini embeddings

Fix Gemini model and Qdrant search

Update backend with fixes

Deploy FastAPI backend for Physical AI Textbook

initial commit